CDK Depict has great structure and substructure functionality for this ... based of SMILES/SMARTS
https://www.simolecule.com/cdkdepict/depict.html

https://www.simolecule.com/cdkdepict/depict/cow/svg?smi=CN1C%3DNC2%3DC1C(%3DO)N(C(%3DO)N2C)C%20caffeine&abbr=off&hdisp=bridgehead&showtitle=false&sma=Cncc(%3DO)&zoom=1.6&annotate=none

@mwang87 what do you mean by "peak annotation provided as part of USI"? That the source of the peaks provides annotations, or that we actually perform the annotation from the peaks?

The former feels a bit neater to me for this tool. In my head, the tool we're making is a way of displaying stuff stored elsewhere....is there a standard format for storing annotations along with MS2 data?

I have a final year UG student doing a project on building a javascript spectral plotting tool and part of that is how to handle annotations...we could just pull him straight into here as it seems a fertile place for suggestions?

I would say it should be from the source, many (but not all) MassBank spectra have this as formulas, not many as fragment structures. RMassBank-created spectra have a consistent format but it's not applied over the whole database.

Or there'd have to be an in silico fragmenter involved I guess and done USI end, but then this should be clear ... (METLIN debate...)

I think it would be neater if people provided the annotation, though I think its not very common as @schymane said, its likely not very common. Perhaps the tools are not super mature at the moment, so we can punt on this for the time being.